Hi, I have a few question's in relation to chapter 19 of the notes. 3.1 Relevance and credibility of past data (page 7 and 8) - Changes in the mix of homogeneous groups within the past data Would an example of this be the a group of people that was grouped together because of say underwriting (all suffered from previous health condition), but then after 10 years further underwriting has been completed and some are not cleared from medical conditions and therefore groups will change? - Changes in the mix of homogeneous groups to which the assumptions apply If I am correct the interpretation of the above point, struggling to see how the second point differs. 3.2 Use of real values (page 7 and 8) Past data for price inflation can be very useful in determining other economic assumptions, as conversion of past economic data into real terms will often remove much of the fluctuation. I don’t really get the second part of this statement regarding conversion of data eliminates fluctuations? For this reason, actuaries often develop a set of assumptions in real terms. To the extent that all the factors affecting future cashflows can be determined relative to price inflation, real parameters are all that are required. Does this mean the financial parameters such as discount rate etc? And price inflation is the benchmark they are derived from so they are in real terms? 3.2 One off Impacts (Page 9) An example of this could be a material change in the value of fixed interest assets as a result of the central bank or government changing short-term interest rates materially, or as a result of a change in the level of government bond issues The first part of statement is ok I think – basically if the short-term interest rate changes, then the coupon payments of a government bond/fixed-interest asset will change. But for the second part regarding the level of government bonds issued – does this mean the price that they are issued at? So you make a big return as if price increases you can sell asset (as you bought at lower price) at increased price? Section 5.1 Margins (page 14) The basis actually chosen for pricing, inclusive of margins, will fix the risk the company will be subject to once the contract is issued at that price. Why does this fix the risk? Does having margins not reduce the risk? I assume basis means the assumptions? Section 5.3 Profit Criterion (page 15) For example, a possible criterion for an insurer is that the net present value of profits emerging from each of its product lines is a predetermined proportion of the distribution costs. Such a criterion reduces the bias towards products with high commission rewards in the distribution system. So basically this method (predetermined proportion) is used to highlight/isolate products with very high commission? How does it work exactly and what is the issue with high commission? Not sure exactly what it means. Many thanks in advance, Darragh
Hi Darragh Section 3.1 queries: Insurance data may be grouped so that policies in a group are not similar in terms of risk characteristics – in other words, groups have heterogeneity and are comprised of a number of subgroups of policy, each with different risk characteristics (ie a number of different homogeneous groups are included). This heterogeneity is not bad in itself, but it presents a risk – the risk that the balance between the homogeneous subgroups changes. This issue is that the past experience data used to set assumptions for current policies will not be appropriate if the thing mentioned in either of the two bullet points you quoted happen. If either of these things happen, the risk characteristics of the group providing experience to set the assumptions are highly likely to be different to those for the group the assumptions are being set for, leading to incorrect / inappropriate assumptions being set. Section 3.2 queries – real values: Nominal parameters can vary a lot over time, but much of this variation is due to inflation variation. If we strip inflation out of our assumptions (by using ‘real’ assumptions) we therefore have assumptions that are easier to set, and that won’t need to vary as much over time, hence the statement about actuaries often using real terms assumptions. The point of the last bit you have quoted is that, if both income and outgo increase in line with inflation, the impact of inflation changes will (largely) net off between assets and liabilities. It’s therefore not necessary to allow for inflation changes in valuing assets and liabilities and we can project and discount both forwards then backwards (respectively) using rates net of inflation. Section 3.2 queries – one off impacts: Regarding the first part of the statement, as we saw in our Day 2 tutorial, the expectations theory says that bond yields will change if future short-term yield expectations change (eg if the central bank or government unexpectedly changes short-term interest rates materially). If yields change in one direction, as we also saw on Day 2, bond prices will move in the opposite direction. Regarding the second part of the statement, supply of bonds will change if the government issue / redeem bonds (ie changes to the level of government bond issues) – supply will increase / reduce respectively … decreasing or increasing, respectively the value (ie price) of bonds. Section 5.1 queries: Yes - basis does mean assumptions. You are correct that having margins will reduce risk, but some risk will always remain and, once the premium for a policy is set, it can’t be change over the term of the policy at least and so risk can’t be further reduced by increasing margins and hence premiums as these are fixed. Section 5.3 queries: The answer to your first question is ‘yes’ – we thereby find policies where commission is disproportionately high compared to profits. The idea would then be to aim to sell less of these and more of the products where the opposite is true. Commission will be a cost to an insurer – the insurer will want to reduce costs all else being equal. I hope that helps.
Hi James, Thanks for the detailed responses really appriciate it. Just last q on section 3.1, th rest is all clear now. 3.1 Relevance and credibility of past data (page 7 and 8) OK just to confirm - an example of a group where the risks are different would be a group of males 50-60. Within the group there are sub-groups - smokers, individuals who've suffered from a stroke, individuals who've been in a serious accident. All of these sub-groups in themselves are homogeneuos but the overall group has heterogeneity? So if over time the number of indiviuals in the sub-group of smokers increase over say 20 years then the balance has changed, and then this past data of males and the sub-groups is not good data to be used to set future assumptions? Just find it difficult to distiguish between: - Changes in the mix of homogeneous groups within the past data and - Changes in the mix of homogeneous groups to which the assumptions apply As they seem pretty similar to me. Thanks, Darragh
Hi Darragh You've got the right idea and your example is a good one. In summary, to use data to set assumptions, the risks underlying the experience data being used need to be consistent with those posed by the policies the assumptions will be used for. With heterogeneity, there's a greater risk that this won't be the case. As per the bullet points you mention, there's two ways this can happen if groups are heterogeneous: 1. If we are using experience from the same group but the balance of the homogeneous groups within this heterogeneous group has changed over time, the average group over the (past) period the data was collected in won't be the same as the group now. 2. If we apply assumptions, then there are changes in the balance of homogeneous groups in the heterogeneous group to which the assumptions are applied, the assumptions may no longer be appropriate as, again, there will be a difference between the group underlying the experience data and the group to which assumptions are being applied. I hope this helps.
Hi James, Apololgies on delayed response. That's all clear now thanks for explaining that in detail. Thanks, Darragh
